Journal of Financial Econometrics Advance Access originally published online on September 28, 2006
Journal of Financial Econometrics 2006 4(4):671-675; doi:10.1093/jjfinec/nbl006
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
© The Author 2006. Published by Oxford University Press. All rights reserved. For permissions, please e-mail: journals.permissions@oxfordjournals.org.
Practitioners’ Corner
a.canopius@cirano.qc.ca
| The first 150 words of the full text of this article appear below. |
A fundamental issue in international portfolio diversification relates to the evolution over time of the correlations and variances of assets. If correlations were to increase in bad times or when markets were highly volatile, the diversification benefits would certainly be compromised when most needed. However, statistical complications arise in addressing the issue. The usual approach to test whether correlation changes during bear period conditions on observed ex post realized low market returns. Yet care must exercised in testing such a proposition, because correlation is a complex function of returns. Boyer, Gibson, and Loretan (1999)
show that conditional correlation is highly nonlinear in the level of returns on which it is conditioned. One cannot conclude that the true correlation is changing simply by looking at the difference between the values obtained conditioning, for example, on low or high values of one variable or both variables. The distribution of conditional correlation under